Age-Sex Pyramid by SG Planning Area from 2000

1.0 Introduction

The aim of this exercise is to design an age-sex pyramid based data visualisation to show the changes of demographic structure of Singapore by age cohort and gender between 2000-2020 at planning area level.The data used is Singapore Residents by Planning Area / Subzone, Age Group, Sex and Type of Dwelling, June 2000-2010 and Singapore Residents by Planning Area / Subzone, Age Group, Sex and Type of Dwelling, June 2011-2020 obtained from Singstat.

I would like to explore the following visualisations:

Animated age-sex pyramid that changes across time for 1 planning area
Age-sex pyramid that changes for 1 planning area in a particular year with tooltip
Age-sex pyramid comparing a few planning area and how the population distribution changes across time for these areas
Check if filter can be applied on planning area and time

2.0 Data Preparation

Install the necessary R packages - ‘tidyverse’, ‘readxl’, ‘knitr’, ‘ggplot2’, ‘plotly’,‘DT’,‘patchwork’,‘gganimate’,‘gifski’,‘gapminder’,‘ggiraph’.

packages = c('tidyverse', 'readxl', 'knitr', 'ggplot2', 'plotly','DT','patchwork','gganimate','gifski','gapminder','ggiraph')

for(p in packages){
  if(!require(p, character.only = T)){
    install.packages(p)
  }
  library(p, character.only = T)
}

2.1 Importing & Joining 2 Dataframe

As the 2000-2010 data and the 2011 - 2020 data are in two separate files, we will first need to import them using read_ccv() and append them together using rbind().

pop_2000to2010<-read_csv("data/respopagesextod2000to2010.csv")
pop_2011to2020 <-read_csv("data/respopagesextod2011to2020.csv")
pop_combined<- rbind(pop_2000to2010,pop_2011to2020)

2.2 Data Preparation

Sum the population after grouping by planning area, year, age group and gender.

Pop_grp <- pop_combined %>%
  group_by(PA, Time, AG,Sex) %>%
  summarise(`Total Pop`=sum(Pop))%>%
  ungroup()

Pop_grp

# A tibble: 43,890 x 5
   PA          Time AG       Sex     `Total Pop`
   <chr>      <dbl> <chr>    <chr>         <dbl>
 1 Ang Mo Kio  2000 0_to_4   Females        4460
 2 Ang Mo Kio  2000 0_to_4   Males          4830
 3 Ang Mo Kio  2000 10_to_14 Females        5520
 4 Ang Mo Kio  2000 10_to_14 Males          5710
 5 Ang Mo Kio  2000 15_to_19 Females        5930
 6 Ang Mo Kio  2000 15_to_19 Males          6260
 7 Ang Mo Kio  2000 20_to_24 Females        7160
 8 Ang Mo Kio  2000 20_to_24 Males          7170
 9 Ang Mo Kio  2000 25_to_29 Females        7750
10 Ang Mo Kio  2000 25_to_29 Males          8260
# ... with 43,880 more rows

Split the age group to starting age and ending age of the age group using separate() function in dplyr package and convert them to numeric.This is to enable the sorting of age group by starting age later.Split the Age group to the starting age and ending age of the age group using separate() function in dplyr package and convert them to numeric.This is to enable the sorting of age group by starting age later.

Pop_grp2<-separate(data=Pop_grp,col = AG,into=c("Starting_Age","Ending_Age"),
                   remove = FALSE, sep = "_to_")
Pop_grp2$Starting_Age<-as.numeric(Pop_grp2$Starting_Age)
Pop_grp2$Ending_Age<-as.numeric(Pop_grp2$Ending_Age)

2.3 Data visualisation

2.3.1 Animated Age-Sex Pyramid For One Planning Area Across Time

Attempt to draw an animated Age-Sex Pyramid by filtering out 1 planning area first.The following chart shows how the population varies in Ang Mo Kio planning area from 2000 to 2020.

Pop_filter<-Pop_grp2 %>%
  filter(PA=="Ang Mo Kio")

ggplot(Pop_filter, aes(y = reorder(AG,Starting_Age), fill= Sex, 
                           x = ifelse(test = Sex =="Males",
                                      yes = -`Total Pop`, no = `Total Pop`)))+
  geom_col(stat="identity")+
  scale_x_continuous(labels = abs, limits = max(Pop_filter$`Total Pop`) * c(-1,1))+  
  labs(title = 'Age-Sex Pyramid for Ang Mo Kio Population in:{as.integer(frame_time)}', x="Population", y ="Age Group")+
  transition_time(Time)+
  ease_aes('linear')

Try adding tooltip using the ggiraph package with data for a particular year and planning area. Geom_col_interactive() and girafe() are used to create this interaction.

Pop_filter<-Pop_grp2 %>%
  filter(PA=="Ang Mo Kio",Time=="2000")

Pop_filter$tooltip<-c(paste0("Planning Area= ",Pop_filter$PA,
                             "\n Gender= ", Pop_filter$Sex,
                             "\n Pop = ", Pop_filter$`Total Pop`))

q<-ggplot(Pop_filter, aes(y = reorder(AG,Starting_Age), fill= Sex, 
                           x = ifelse(test = Sex =="Males",
                                      yes = -`Total Pop`, no = `Total Pop`)))+
  geom_col_interactive(tooltip=Pop_filter$tooltip)+
  scale_x_continuous(labels = abs, limits = max(Pop_filter$`Total Pop`) * c(-1,1))+  
labs( title = 'Age-Sex Pyramid for Ang Mo Kio Population in 2000',
      x="Population", y ="Age Group")  

girafe(
  ggobj = q,
  width_svg = 8,
  height_svg = 8*0.618
)

2.3.3 Animated Age-Sex Pyramid For A Few Planning Area Across Time

Filter raw data for the selected planning areas that we are keen to explore instead of drawing the age-sex pyramid for all planning areas to prevent cluttering of graphs. In this case, I have selected Ang Mo Kio & Bishan vs Punggol to observe how the population varies in more mature areas and less mature area across the years. facet_wrap() is used to show the charts split by the selected planning areas.

Pop_filter2<-Pop_grp2 %>%
  filter(PA=="Ang Mo Kio"|PA=="Bishan"|PA=="Punggol")

r<-ggplot(Pop_filter2, aes(y = reorder(AG,Starting_Age), fill= Sex, 
                           x = ifelse(test = Sex =="Males",
                                      yes = -`Total Pop`, no = `Total Pop`)))+
  geom_col(stat="identity")+
  scale_x_continuous(labels = abs, limits = max(Pop_filter2$`Total Pop`) * c(-1,1))+  
  labs(title = 'Age-Sex Pyramid for SG Population in:{as.integer(frame_time)}', 
       x="Population", y ="Age Group")+
  transition_time(Time)+
  ease_aes('linear')

r+theme(aspect.ratio = 1)+facet_wrap(~PA, scales = "free_x")

2.4 Challenges & Learnings

Attempted to use crosstalk together with plotly to design an age-sex pyramid that changes based on the year and planning area filter. However, the filter function only shows the position of data point after all data points are plotted instead of filtering the raw data to draw a new plot.

After researching online,this could be due to the inherent limitations of crosstalk that “Crosstalk currently only works for linked brushing and filtering of views that show individual data points, not aggregate or summary views (where “observations” is defined as a single row in a data frame)."

It is definitely more challenging to use R in creating the filters vs creating similar visualisations using Tableau.

library(crosstalk)


SharedPop<-SharedData$new(Pop_grp2)

p<-ggplot(SharedPop, aes(y = reorder(AG,Starting_Age), fill= Sex, 
                           x = ifelse(test = Sex =="Males",
                                      yes = -`Total Pop`, no = `Total Pop`)))+
  geom_col()+
  scale_x_continuous(labels = abs)+  
  labs(title = 'Age-Sex Pyramid for SG Population', x="Population", y ="Age Group")

bscols(widths=c(2,NA,NA),
      list(
        filter_select("PA","Planning Area", SharedPop,~PA,multiple = FALSE),
        filter_select("Time","Year", SharedPop,~Time,multiple = FALSE)
        ), 
     ggplotly(p)
       )

Planning Area

Year

Age-Sex Pyramid by SG Planning Area from 2000 - 2020